W3C home > Mailing lists > Public > www-style@w3.org > November 2009

Re: [CSS21] margin computed value

From: Andrey Mikhalev <amikhal@abisoft.spb.ru>
Date: Mon, 9 Nov 2009 23:34:53 +0300 (MSK)
To: "L. David Baron" <dbaron@dbaron.org>
cc: www-style@w3.org
Message-ID: <Pine.LNX.4.63.0911092301520.25651@master.abisoft.spb.ru>
On Mon, 9 Nov 2009, L. David Baron wrote:

> On Monday 2009-11-09 21:14 +0300, Andrey Mikhalev wrote:
>> On Mon, 9 Nov 2009, L. David Baron wrote:
>>> On Monday 2009-11-09 18:07 +0300, Andrey Mikhalev wrote:
>>>> in "8.3 Margin properties", computed value for margin-* claimed as:
>>>> "the percentage as specified or the absolute length"
>>> I think this should say "the percentage or auto as specified, or the
>>> absolute length".
>> mozilla POV :)
> Not really.  Mozilla used to inherit the old definition of computed
> value, but it was a lot of extra work, and getting rid of that extra
> work was the whole point of the change that introduced used values.
> The point of that change was that inheritance would never depend on
> layout-dependent computations, which it would if margin:auto has to
> be inherited as a number.  It would be extremely silly to remove
> that dependency for all cases but one.

well, "Computed values are processed as far as possible without formatting 
the document".
resolving margin:auto needs layout-dependent computations only for
in-flow block-level and absolutely positioned elements
(10.3.3/4, 10.3.7/8)
there's no reason to defer resolving of 'auto' to 0 in all other cases.
hence keeping 'auto' _always_ 'as specified' is mozilla POV.

(not advocate it, just notice)

> -David
Received on Monday, 9 November 2009 20:35:33 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:07:40 UTC